108 Mr. H. L. Callendar on the 



pyrometer-coil, so that the change of its resistance between 0° 

 and 100° C. may be equal to 1, 10, or 100 units. The divisions 

 of the bridge- wire will then represent degrees of temperature 

 on the platinum scale, and the temperature can be deduced 

 from the observed resistance by simply subtracting its zero 

 value. 



The arrangement, however, will work equally well with an 

 arbitrary scale. The really important point is that an obser- 

 vation of temperature is reduced to a single scale-reading, 

 and that it is independent of the length of stem immersed. 



The chief errors to be apprehended with this form of appa- 

 ratus are changes in the screw and plug contacts, especially 

 the latter. Instead of the ordinary plug for short-circuiting 

 the terminals of a resistance-coil, I prefer to use a flat nut 

 screwing down on to two flat plates. I find that better con- 

 tact may thus be secured, and, what is more important, that 

 the contact-surfaces may be more easily kept clean and good. 

 It is almost impossible to keep a plug-hole free from grit and 

 grease, even if the plug fits perfectly ; whereas the screw-nut 

 can be kept screwed down when not in use, and the flat sur- 

 faces are easily cleaned if necessary. Moreover 'the wedge- 

 like action of the plugs almost invariably loosens the blocks 

 between which they are inserted, so that taking out one plug 

 alters the contacts of its neighbours. This cannot occur with 

 the screw-nuts. Among minor advantages it may be men- 

 tioned that the screw-nuts, not being loose, are not so liable 

 to get lost or soiled. 



Perhaps the best way to make the sliding-contact at H is 

 to connect the galvanometer to a fixed wire of the same 

 material as FK and stretched parallel to it. Connexion is 

 made between the two wires by pressing down a button carry- 

 ing a short cross-wire fixed to a spring on the lower side of 

 the sliding contact-piece. The advantages of this arrange- 

 ment are that the wire from the galvanometer is not attached 

 to the sliding-piece itself but to a fixed binding-screw, and 

 that there is less danger of thermoelectric effects being 

 produced by the warmth of the hand at H. 



The current used in measuring the resistance must not be 

 too large or the fine wire will be heated. It appears that a 

 platinum wire of 0*015 centim. ( # 006 inch) diam., which is a 

 convenient size to use for these thermometers, will take a 

 current of one hundredth of an ampere when cooled by air 

 contact, without heating much more than one hundredth of a 

 degree C. If a higher order of accuracy is desired, the 

 current-heating must be measured and allowed for. 



The degree of sensitiveness and accuracy attainable with 



